Small-scale public space digital multi-dimensional vitality assessment technical method

Abstract

The invention relates to the technical field of urban planning and smart cities, and particularly discloses a small-scale public space digital multi-dimensional vitality assessment technical method. According to the method, an automatic evaluation system based on computer vision and data fusion is constructed, a 3M-VA model is adopted, and public space vitality is quantified from three dimensions of aggregation, stability and complexity. The system comprises a data acquisition module, a processing and algorithm module and a statistical analysis module, behavior characteristics such as the number of pedestrians, trajectory length and residence time are automatically extracted through target detection, multi-target tracking and trajectory analysis algorithms, and multi-dimensional vitality indexes are integrally calculated; meanwhile, a statistical association model of environment variables and vitality dimensions is established by processing form, vision, facility and other multi-source environment data. The method overcomes the defects that a traditional method is high in subjectivity, single in dimension and insufficient in precision, efficient, objective and refined vitality evaluation is achieved, and a scientific basis and operable decision support can be provided for public space optimization design.

Description

Technical Field

[0001] This invention relates to the fields of urban planning and smart city technology, specifically to a digital, multi-dimensional, and automated assessment method for the vitality of small-scale urban public spaces based on computer vision and data fusion. Background Technology

[0002] Small-scale public spaces are crucial nodes in the urban fabric, and their vitality directly impacts the daily experience of community residents and the quality of the urban microenvironment. Traditional vitality assessment methods rely on manual observation and questionnaires, which suffer from significant drawbacks such as high subjectivity, high cost, low efficiency, discontinuous data, and difficulty in capturing behavioral complexity. Although recent studies have attempted to utilize data technologies such as mobile phone signaling and Wi-Fi probes, these methods pose risks of privacy breaches, data update delays, and low spatial resolution, making it difficult to accurately reflect individual behavioral patterns within small-scale spaces.

[0003] In existing technologies, assessment methods based on single indicators (such as pedestrian flow / crowd density) cannot fully reflect the rich connotations of spatial vitality. A truly vibrant small-scale space not only needs to attract crowds (high aggregation), but also requires people to be willing to stay there for extended periods (high stability) and engage in diverse activities (high complexity). Therefore, there is an urgent need for an assessment method that can objectively, efficiently, multidimensionally, and comprehensively reflect the vitality of small-scale public spaces.

[0004] Existing patents CN202510607179.5 and CN202410080235.X propose a scheme for counting pedestrian traffic in public places based on video surveillance. This scheme collects pedestrian traffic data in a specific area using video equipment and then counts the pedestrian traffic. However, this scheme only focuses on a single aggregation indicator and cannot effectively assess the high stability and high complexity of spatial vitality, making it difficult to solve the problem of multi-dimensional vitality assessment.

[0005] In terms of spatial applications, patent CN202210511466.2 uses Bluetooth wireless communication to determine crowd density. It generates a judgment matrix by detecting the Bluetooth RSSI value and signal angle of arrival of the mobile phone and processes it to obtain density parameters, which is suitable for underground or indoor spaces. Patent CN202510709441.7 uses thermal imaging cameras, WiFi probes and infrared sensors to build a perception network, obtain various data to generate heat distribution maps and model the spatiotemporal correlation of crowd flow, but it is mainly aimed at AI exhibition hall scenarios. Although patent CN201710514073.6 can obtain Bluetooth / Wi-Fi signals and sensor data through smartphones to generate urban population density and flow maps, it is mainly aimed at the macro level of urban areas and lacks specificity for the complex multi-dimensional vitality assessment of small-scale public spaces, making it difficult to accurately reflect the vitality of small-scale public spaces.

[0006] Therefore, developing a multi-dimensional vitality assessment technology that is designed specifically for small-scale public spaces and is fully digitalized throughout the entire process has significant application value. Summary of the Invention

[0007] The core of this invention lies in providing a digital multidimensional vitality assessment technology and method for small-scale public spaces. Through an innovative technical framework (3M-VA model), a three-dimensional vitality assessment model of "clustering-stability-complexity" is constructed. Utilizing video data and computer vision technology, it achieves automated, refined, and multidimensional quantification of the vitality of small-scale public spaces, generating comparable and traceable digital vitality assessment results, and identifying environmental driving factors, as shown in the appendix. Figure 1 As shown.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: First, a small-scale, multi-dimensional, automated vitality assessment system based on computer vision is constructed, characterized by including: a data acquisition module, a data processing and algorithm module, and a statistical analysis module. 1. Data Acquisition Module: This module mainly consists of two parts: vitality data acquisition and built environment data acquisition. It is configured to acquire video data of the target small-scale public space; and to process multi-source spatial data to quantify environmental variables related to vitality, including: Morphological data, such as area (AREA) and width-to-depth ratio (W / D), are obtained through the OpenStreetMap data platform and field survey data.

[0009] Visual data, such as green view ratio (GVR) and sky visibility (SVF), are obtained by taking small-scale panoramic and fisheye photographs of public spaces on-site.

[0010] Facility data, such as Facility Richness (RF), is quantified through on-site audits and using a tiered system (e.g., a 5-point scale).

[0011] 2. Data Processing and Algorithm Module: This module mainly consists of two parts: vitality data processing and built environment data processing. These are configured to run the 3M-VA model, as shown in the attached diagram. Figure 1 As shown; the vitality data processing and algorithm module includes a function for calculating clustered vitality (V). A The system includes a target detection unit (unit 1), a multi-target tracking unit (unit 2) for generating pedestrian trajectories, and a stability vitality (V) calculation unit. S ) and complexity vitality (V C The trajectory analysis unit (unit 3) and the unit for calculating total vitality (V) total The vitality integrated computing unit (unit 4): Object detection unit: Based on a convolutional neural network (CNN) model, it processes video frames, automatically identifies and frames users in the space, and outputs the number of people at each time point to calculate the aggregation vitality (V). A ).

[0012] Multi-target tracking unit: Based on discriminative correlation filters (such as the DCF-CSR algorithm), it extracts features such as color names and histogram of gradient directions (HOG), matches pedestrian identities across frames, and generates continuous motion trajectories.

[0013] Trajectory Analysis Unit: Analyzes the generated trajectory and quantifies stability vitality (V) by calculating the dwell time of the user's trajectory. S ), and by calculating the ratio of the actual length of the trajectory to the straight-line distance between the start and end points (L) trajectory / D OD ) to quantify complexity vitality (V) C ).

[0014] Vitality Integrated Calculation Unit: Based on the formula (V) A × V S × V C Calculate the total vitality value (V) total ).

[0015] The built environment data processing includes units for calculating morphological features, visual features, and facility features: Morphological feature units: Area (AREA), width-to-depth ratio (W / D), etc. are calculated by processing data obtained from OpenStreetMap and field surveys using GIS software (such as ArcGIS).

[0016] Visual feature unit: Green view ratio (GVR) is calculated by segmenting panoramic image pixels, and sky visibility (SVF) is calculated by processing fisheye photos using Rayman software.

[0017] Facility Feature Unit: Facility richness (RF) is quantified through on-site audits and using a tiered (e.g., 5-point) system.

[0018] 3. Statistical Analysis Module: Configured for (e.g., using SPSS) performing stepwise multiple linear regression analysis, establishing a statistical model between built environment characteristics (independent variables) and vitality values ​​of each dimension (dependent variables), and identifying significantly correlated driving factors. Subsequently, based on the established evaluation system, a computer vision-based automated evaluation method for the multidimensional vitality of small-scale public spaces was developed, characterized by the following steps: Step 1: Selection of Assessment Objects and Data Collection. Small-scale public space vitality assessment objects are selected based on criteria such as proximity to high-density residential areas, similar external environments, and diverse internal characteristics; video data and built environment data (panoramic views, fisheye photos, facility lists, etc.) of the assessment objects are collected.

[0019] Step 2: Video Preprocessing and Model Calculation. Frame extraction is performed on the video; the video frames are input into the 3M-VA model, and object detection and multi-object tracking algorithms are used for processing to calculate the number of people and extract pedestrian trajectories. Based on the number of people and trajectories, the aggregation vitality (V) is calculated. A Stability and vitality (V) S ) and complexity vitality (V C ), and further synthesize total vitality (V total ).

[0020] Step 3: Quantify environmental variables. Process the collected environmental data and calculate its area (AREA), width-to-depth ratio (W / D), green view ratio (GVR), sky visibility (SVF), facility richness (RF), etc.

[0021] Step 4: Correlation Analysis. Stepwise multiple linear regression analysis is performed between the calculated vitality indicators and the quantified environmental variables to reveal key driving relationships.

[0022] Step 5: Results Output and Application. Generate a vitality space distribution map and a driving factor analysis report to provide evidence for the precise design of small-scale public spaces.

[0023] Beneficial effects 1. Multi-dimensional Automated Assessment: This approach automates and digitizes the assessment of vitality (clustering, stability, complexity) in small-scale public spaces, overcoming the limitations of traditional methods that suffer from single-dimensionality and low efficiency. In exploring vitality dimensions, it can identify dimensions that cannot be determined subjectively, such as the vitality level of the complexity dimension based on trajectory feature measurements.

[0024] 2. High data accuracy: By utilizing video data and computer vision algorithms, high accuracy is achieved at a small scale, far surpassing big data methods such as mobile signaling.

[0025] 3. Clear Mechanisms Revealed: Through rigorous statistical models, the differential driving effects of different built environment characteristics on specific vitality dimensions are clearly revealed, surpassing previous studies that only focused on overall vitality.

[0026] 4. Strong design guidance: It reveals specific and transferable design insights, providing planners with precise and actionable optimization levers.

[0027] 5. High scalability: After adaptive adjustments, the system and method can be applied to assess the spatial vitality of various small-scale public open spaces, including city squares, streets, and parks.

[0028] 6. To establish a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the systems and methods described above. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of the overall research framework of the system described in this invention, including a data acquisition module and a data processing and algorithm module for two parts: the vitality of small-scale public spaces and the characteristics of the built environment, and a statistical analysis module for exploring the correlation between the vitality of small-scale public spaces and the characteristics of the built environment.

[0030] Figure 2 The image shows a plan of Nanjing Hydrangea Park and 12 small-scale public spaces selected from it, according to a specific embodiment of the present invention.

[0031] Figure 3 This describes the distribution of total vitality levels at each site in the specific embodiments of the present invention. Detailed Implementation

[0032] The following uses the Hydrangea Park in Nanjing, China as an example to illustrate the specific implementation of the present invention.

[0033] Step 1: Selection of evaluation subjects and data collection.

[0034] Selection of assessment objects: Based on the criteria, 12 representative waterfront activity sites were selected as the vitality assessment objects, all of which are small-scale public spaces.

[0035] Data Acquisition: During the same time period on a summer weekend afternoon, a DJI Mavic 2 Pro drone was used to collect a total of 1200 minutes of video data according to a preset dual-flight plan. Simultaneously, built environment morphology data of the site was collected using ArcGIS and on-site mapping, including area (AREA), width-to-depth ratio (W / D), and distance to water bodies (D). water Panoramic images and fisheye photos were taken on-site as data sources for Green View Rate (GVR) and Sky Visibility (SVF), and facility audits were conducted to obtain facility data for the site, specifically facility richness (RF).

[0036] Step 2: Video preprocessing and model calculation.

[0037] The video data is processed by frame extraction and preprocessing, then input into the 3M-VA model. Object detection and tracking algorithms are used to process the video data and calculate the aggregation vitality (V) of each site. A Stability and vitality (V)S Complexity vitality (V) C ) and total vitality (V) total )value.

[0038] Step 3: Establish environmental variable quantification.

[0039] Calculate area (AREA), width-to-depth ratio (W / D), and distance from water body (D) in ArcGIS. water Green view ratio (GVR) was calculated by segmenting panoramic images using Photoshop; sky visibility (SVF) was calculated by analyzing fisheye photos using Rayman software; and facility richness (RF) was assessed through field surveys using a 5-point scale.

[0040] Step 4: Correlation analysis.

[0041] SPSS was used to perform stepwise multiple linear regression analysis on data from 12 sites (N=240), including various built environment variables and aggregation activity (V). A Stability and vitality (V) S Complexity vitality (V) C ) and total vitality (V) total The relationship between ).

[0042] Step 5: Output and application of results.

[0043] The distribution of total vitality levels is as follows Figure 3 As shown in the table, the vitality status of each site can be observed. The statistical analysis results of the driving factors are shown in Table 1. The analysis results indicate that total vitality (V...) total The main factors are facility richness (RF), width-to-depth ratio (W / D), sky visibility (SVF), and distance from water (D). water Four variables had a significant impact (adjusted R² = 0.727). The driving factors varied across dimensions: aggregation activity (V... A ) is determined by facility abundance (RF) and distance from water body D water Drive; Stability and Vitality (V) S The factors driving this are facility richness (RF), width-to-depth ratio (W / D), and sky visibility (SVF); complexity vitality (V) C It was found that the vitality was driven solely by facility richness (RF). Furthermore, the positive effect of green view ratio (GVR) on vitality diminished after exceeding 42.5%.

[0044] Table 1. Analysis of built environment factors driving the vitality level of 12 sites in Hydrangea Park (N=240)

[0045] *p<0.1, **p<0.05 The above embodiments verify the effectiveness and practicality of the method of the present invention. The present invention provides a standardized and operable technical process that enables scientific, objective, and multi-dimensional digital assessment of the vitality of small-scale public spaces, providing important technical support for improving the quality of urban micro-spaces.

Claims

1. Constructing a small-scale, multi-dimensional, automated vitality assessment system based on computer vision, characterized in that: include: Data acquisition module, data processing and algorithm module, and statistical analysis module: Data Acquisition Module: This module mainly consists of two parts: vitality data acquisition and built environment data acquisition, which are configured to collect video data of the target small-scale public space. Configuration for processing multi-source spatial data to quantify environmental variables related to vitality, including: Morphological data, such as area and aspect ratio, are obtained through the OpenStreetMap data platform and field survey data; Visual data, such as green view rate and sky visibility, are obtained by taking small-scale panoramic photos and fisheye photos of public spaces on-site. Facility data: such as the richness of facilities, which is quantified through on-site audits and by using a tiered system, such as a 5-point system; Data Processing and Algorithm Module: This module mainly consists of two parts: vitality data processing and built environment data processing, each configured to run the 3M-VA model. The vitality data processing and algorithm module includes a target detection unit for calculating clustered vitality, a multi-target tracking unit for generating pedestrian trajectories, a trajectory analysis unit for calculating stable and complex vitality, and a vitality integration calculation unit for calculating total vitality. Target detection unit: Based on the convolutional neural network model, it processes video frames, automatically identifies and frames users in the space, and outputs the number of people at each time moment for calculating cluster activity. Multi-target tracking unit: Based on discriminative correlation filters, such as the DCF-CSR algorithm, it extracts features such as color and gradient direction histogram, matches pedestrian identities across frames, and generates continuous motion trajectories; Trajectory Analysis Unit: Analyzes the generated trajectory, quantifies stability vitality by calculating the dwell time of the user's trajectory, and quantifies complexity vitality by calculating the ratio of the actual length of the trajectory to the straight-line distance between the start and end points; Vitality Integration Calculation Unit: Calculates the total vitality value based on the product of clustered vitality, stable vitality, and complex vitality; The built environment data processing includes units for calculating morphological features, visual features, and facility features: Morphological feature units: Area, aspect ratio, etc. are calculated by processing data obtained from OpenStreetMap and field surveys using GIS software; Visual feature unit: calculates green visibility by segmenting panoramic image pixels and calculates sky visibility by processing fisheye photos using Rayman software; Facility characteristic unit: Facility richness is quantified through on-site audits and by using a tiered system, such as a 5-point system; Statistical Analysis Module: Configured to perform stepwise multiple linear regression analysis, establish a statistical model between built environment characteristics and vitality values ​​of various dimensions, and identify significantly correlated driving factors.

2. According to the system described in claim 1, a method for automated assessment of multi-dimensional vitality in small-scale public spaces based on computer vision is developed, characterized in that... Includes the following steps: Step 1: Selection of assessment objects and data collection; Based on criteria such as proximity to high-density residential areas, similar external environment, and diverse internal characteristics, small-scale public space vitality assessment objects are selected; Video data and built environment data of the assessment objects are collected, including panoramic images, fisheye photos, facility lists, etc. Step 2: Video preprocessing and model calculation; perform frame extraction on the video; input the video frames into the 3M-VA model, process them using target detection and multi-target tracking algorithms, calculate the number of people, and extract pedestrian trajectories. Based on the number of people and trajectories, calculate the clustering vitality, stability vitality, and complexity vitality, and further synthesize the total vitality. Step 3: Establish environmental variable quantification; process the collected environmental data and calculate its area, width-to-depth ratio, green view rate, sky visibility, facility richness, etc. Step 4: Correlation analysis; Stepwise multiple linear regression analysis is performed on the calculated vitality index and the quantified environmental variables to reveal the key driving relationships; Step 5: Output and Application of Results; Generate a distribution map of vitality spaces and an analysis report of driving factors to provide evidence for the precise design of small-scale public spaces.

3. The system and method according to claims 1-2, characterized in that, The system and method described are applied to assess the spatial vitality of various small-scale public open spaces, including city squares, streets, and parks.

4. A computer-readable storage medium having a computer program stored thereon, the program being executed by a processor to implement the steps of the method as described in any one of claims 1 to 3.

Images